Re: [avr-chat] issues with custom bootloader and ATMega32

[Alexander Krause]

Hi again,


this week I found the issue.

I've confused the bootloader address again :-(

2048bytes weren't bytes but words. That would be word address 0x3800 and
since gcc uses bytes I have to multiply by 2. So that means, that my
boot address is at 0x7000 and not 0x1C00.

*sighs* I've had the same issue about 1 year ago... so hopefully that
was the last time I confused it.

Regards,
Alex

On Di, 2013-10-29 at 13:46 +0100, Alexander Krause wrote:
> Hi all,
> 
> I started to write a bootloader for RS485/ModBus and am stuck for quite
> a while now.
> 
> In general I've three issues.
> 
> First of all, it seems like my IRQ's are not working when I put the code
> into the boot section.
> 
> I've choosen 2048 bytes bootsize, which will start at byte address
> 0x3800 equals word address 0x1C00 .
> Fuses are:
> MCU_LFUSE=0xBE
> MCU_HFUSE=0xD0
> 
> 
> My program starts like this:
> int __attribute__((noreturn)) main(void) {
>       MCUCR = (1 << IVCE);
>       MCUCR = (1 << IVSEL);
> ...
> 
> So that should work, but It doesn't :-(
> 
> Well, since I've to shorten my code as far as possible I simply decided
> to use polling instead of IRQs (saves me the cli, sbi and some other
> things ~100bytes less code).
> 
> This raises the question if I can short the bootloaders interrupt table
> somehow? At the moment it looks like this:
>     1c00:     4f c0           rjmp    .+158           ; 0x1ca0 <__ctors_end>
>     1c02:     00 00           nop
>     1c04:     5d c0           rjmp    .+186           ; 0x1cc0 
> <__bad_interrupt>
>     1c06:     00 00           nop
>     1c08:     5b c0           rjmp    .+182           ; 0x1cc0 
> <__bad_interrupt>
>     1c0a:     00 00           nop
>     1c0c:     59 c0           rjmp    .+178           ; 0x1cc0 
> <__bad_interrupt>
>     1c0e:     00 00           nop
>     1c10:     57 c0           rjmp    .+174           ; 0x1cc0 
> <__bad_interrupt>
>     1c12:     00 00           nop
>     1c14:     55 c0           rjmp    .+170           ; 0x1cc0 
> <__bad_interrupt>
>     1c16:     00 00           nop
>     1c18:     53 c0           rjmp    .+166           ; 0x1cc0 
> <__bad_interrupt>
>     1c1a:     00 00           nop
>     1c1c:     51 c0           rjmp    .+162           ; 0x1cc0 
> <__bad_interrupt>
>     1c1e:     00 00           nop
>     1c20:     4f c0           rjmp    .+158           ; 0x1cc0 
> <__bad_interrupt>
>     1c22:     00 00           nop
>     1c24:     4d c0           rjmp    .+154           ; 0x1cc0 
> <__bad_interrupt>
>     1c26:     00 00           nop
>     1c28:     4b c0           rjmp    .+150           ; 0x1cc0 
> <__bad_interrupt>
>     1c2a:     00 00           nop
>     1c2c:     49 c0           rjmp    .+146           ; 0x1cc0 
> <__bad_interrupt>
>     1c2e:     00 00           nop
>     1c30:     47 c0           rjmp    .+142           ; 0x1cc0 
> <__bad_interrupt>
>     1c32:     00 00           nop
>     1c34:     45 c0           rjmp    .+138           ; 0x1cc0 
> <__bad_interrupt>
>     1c36:     00 00           nop
>     1c38:     43 c0           rjmp    .+134           ; 0x1cc0 
> <__bad_interrupt>
>     1c3a:     00 00           nop
>     1c3c:     41 c0           rjmp    .+130           ; 0x1cc0 
> <__bad_interrupt>
>     1c3e:     00 00           nop
>     1c40:     3f c0           rjmp    .+126           ; 0x1cc0 
> <__bad_interrupt>
>     1c42:     00 00           nop
>     1c44:     3d c0           rjmp    .+122           ; 0x1cc0 
> <__bad_interrupt>
>     1c46:     00 00           nop
>     1c48:     3b c0           rjmp    .+118           ; 0x1cc0 
> <__bad_interrupt>
>     1c4a:     00 00           nop
>     1c4c:     39 c0           rjmp    .+114           ; 0x1cc0 
> <__bad_interrupt>
>     1c4e:     00 00           nop
>     1c50:     37 c0           rjmp    .+110           ; 0x1cc0 
> <__bad_interrupt>
> 
> 
> Can't I simply tell GCC to not use/generate such a table since there'll
> be no IRQ anyway.
> 
> 
> The last and biggest issue is that I'm not able to write via the SPM
> features.
> 
> I tried this with a minimalistic piece of code:
> void spm_write_page(uint32_t page) {
>       uint8_t i;
>       
>       eeprom_busy_wait();
>       boot_page_erase(page);
>       boot_spm_busy_wait();      // Wait until the memory is erased.
> 
>       for (i=0; i<SPM_PAGESIZE; i+=2) {
>               boot_page_fill(page + i, i);
>       }
> 
>       boot_page_write(page);     // Store buffer in flash page.
>       boot_spm_busy_wait();       // Wait until the memory is written.
> 
>       // Reenable RWW-section again. We need this if we want to jump back
>       // to the application after bootloading.
> 
>       boot_rww_enable();
> }
> 
> int __attribute__((noreturn)) main(void) {
>       spm_write_page(1);
>       while(1);
> }
> 
> However, when I dump my Flash there is no valid data in the first page:
> 0000000 ffff ffff ffff ffff ffff ffff ffff ffff
> *
> 0001c00 c04f 0000 c05d 0000 c05b 0000 c059 0000
> 0001c10 c057 0000 c055 0000 c053 0000 c051 0000
> 0001c20 c04f 0000 c04d 0000 c04b 0000 c049 0000
> 0001c30 c047 0000 c045 0000 c043 0000 c041 0000
> 0001c40 c03f 0000 c03d 0000 c03b 0000 c039 0000
> 
> 
> 
> Thanks a lot for reading. I'd be very grateful for any help.
> 
> 
> Regards,
> Alex
> 
> 
> _______________________________________________
> AVR-chat mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/avr-chat
>